Announcement

Collapse
No announcement yet.

EXT4 Changes Land For Linux 5.9 With Block Allocator Performance Work

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • EXT4 Changes Land For Linux 5.9 With Block Allocator Performance Work

    Phoronix: EXT4 Changes Land For Linux 5.9 With Block Allocator Performance Work

    With all the Linux 5.9 kernel changes you may have noticed no major EXT4 file-system pull request was submitted during the kernel merge window the prior two weeks. Fortunately, the EXT4 work has now been sent out and Linus Torvalds honored the late changes for this widely-used Linux file-system...

    Phoronix, Linux Hardware Reviews, Linux hardware benchmarks, Linux server benchmarks, Linux benchmarking, Desktop Linux, Linux performance, Open Source graphics, Linux How To, Ubuntu benchmarks, Ubuntu hardware, Phoronix Test Suite

  • #2
    will be interesting to see how long Fedora default to BTRFS , my guess is, it wont be long

    Comment


    • #3
      Why is that?

      Comment


      • #4
        Originally posted by Spam View Post
        Why is that?
        Because 'feelings'.

        Comment


        • #5
          just curious: do you have to reformat the disk for new feature to be available?

          Comment


          • #6
            Typically new Ext4 features only were applicable to new data without reformat but you still could enable them, not sure if this will be different.

            Comment


            • #7
              Originally posted by Anvil View Post
              will be interesting to see how long Fedora default to BTRFS , my guess is, it wont be long
              Doesn't need guessing. Fedora is targeting Fedora 33 for default of BTRFS. https://fedoraproject.org/wiki/Changes/BtrfsByDefault

              Comment


              • #8
                But is swapfile on ext4 fixed yet?

                Comment


                • #9
                  Originally posted by set135

                  Probably you were thinking of btrfs, which allows swapfiles as of kernel 5.0 (with some caveats)
                  No, it broke with 5.7.

                  Comment


                  • #10
                    Originally posted by set135

                    Sorry for the assumption. After a little looking, what I see is that swapon is rejecting swapfiles the kernel thinks are created with 'holes', which is a filesystem space optimization where space is not allocated for lots of contiguous nulls in a file. The complaints I saw seemed to relate to swapfiles created with fallocate(1), and were fixed by 'the recommended method' of using dd. I have always used dd since the dark ages for this purpose, so my test worked fine. I just tested with fallocate on kernel 5.8.2 and it rejected it with a log saying 'swapon: swapfile has holes'.

                    So, just to be clear, for me on kernel 5.8.2 a swapfile made with 'fallocate -l 1GB swapblob' is not accepted, but one created with 'dd if=/dev/zero of=swapblob bs=1024k count=1024' works with ext4.
                    Could have sworn I tried dd as well and that failed back when 5.7 was first released, seems to work now though /shrug

                    Comment

                    Working...
                    X